International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019
p-ISSN: 2395-0072
www.irjet.net
AI BASED TRAFFIC SIGNAL CONTROL SYSTEM
Sanchal Ramteke1, Prof. B. B. Gite2
1Student,
Department of Computer Engineering, Sinhgad Academy of Engineering, Savitribai Phule Pune
University, Pune, Maharashtra.
2Head of Department, Department of Computer Engineering, Sinhgad Academy of Engineering, Savitribai Phule
Pune University, Pune, Maharashtra.
----------------------------------------------------------------------***---------------------------------------------------------------------
Abstract - The control system of the traffic light is mainly
used to monitor and control the flow of vehicles through the
intersection of roads. The main purpose of our system is the
smooth movement of cars along the transport line. Integrating
a system with multiple traffic lights into an existing system is a
complex task. The existing system does not control the flow of
the vehicle to the node. There is no common abstraction
between traffic light systems, vehicle deviation, high red light
delay, vehicle drift, accident, emergency vehicle and pedestrian
crossing. The existing system leads to congestion. We
recommend an Arduino based system with image processing
for measuring traffic density. At the traffic light, an image
processing system is used, in which the Arduino regulates the
synchronization of the signal based on the counting of vehicles.
Controls the next signal based on the previous display. The
portable device is designed to solve urgent problems stuck on
crowded roads. MATLAB is software that has a number of
image processing functions.
Key Words: Traffic light system, Arduino, Image
processing, Camera, Vehicle counting, traffic density.
1. INTRODUCTION
A traffic light system developed since 1912 to control traffic
at intersections, pedestrian crossings and other areas. Traffic
jams are increasing day by day, so we have to face many
problems. Due to the large volume of vehicles, the lack of
infrastructure and distribution systems is the main cause of
traffic congestion. Traffic lights are red, blue and green. The
green light signal is used in the indicated direction; the
yellow light signal is used to warn cars of a short stop and a
red light prohibiting movement.
These days, many nations suffer the ill effects of the traffic
congestion issues that influence the transportation system in
urban communities and cause genuine difficulty. Despite
supplanting traffic officers and flagmen via programmed
traffic frameworks, the advancement of the overwhelming
congested driving conditions is as yet a noteworthy issue to
be confronted, particularly with different intersection hubs.
The quick increment in the number of autos and the always
rising number of street clients are not joined by advanced
frameworks with adequate assets. Halfway arrangements
were offered by developing new streets, executing flyovers
and sidestep streets, making rings, and performing streets
restoration.
© 2019, IRJET
|
Impact Factor value: 7.34
|
The traffic issue is extremely muddled because of the
contribution of assorted parameters. In the first place, the
traffic flow depends upon the time where the traffic top
hours are for the most part early in the day and toward the
evening; on the times of the week where ends of the week
uncover least load while Mondays and Fridays by and large
show thick traffic arranged from urban communities to their
edges and in turn around bearing individually; and time as
occasions and summer. Also, the current traffic light system
is executed with hard-coded defers where the lights change
schedule slots are settled routinely and don't rely upon
continuous traffic flow. The third point is worried about the
condition of one light at a crossing point that impacts the
stream of movement at neighboring convergences.
Additionally, the traditional traffic system does not think
about the situation of mischance, road works, and
breakdown vehicle that compound movement blockage.
Furthermore, a pivotal issue is identified with the smooth
movement through crossing points of crisis (Emergency)
vehicles of higher needs, for example, ambulances, rescue
vehicles, fire brigade, police, and VIP people that could stall
out in the crowd. At long last, the walkers (pedestrians) that
cross the paths also modify the traffic system.
The traffic light system should be moved up to fathom or
upgraded the extreme traffic congestion, mitigate
transportation inconveniences, diminish traffic volume and
holding up time, limit by and large travel time, advance
vehicles security and effectiveness, and grow the advantages
in health, economic, and natural segments. This paper
proposes a simple, minimal effort, and constant savvy traffic
light system that means to defeat numerous deformities and
enhance traffic administration. The system depends on
Arduino microcontroller and PC camera, which uses for
continuous capturing the vehicle images on roads and
perform the action to solve the congestion problem in traffic
light system.
2. LITERATURE SURVEY
[1] Introduces a two-stage processing method: vehicle
detection and vehicle detection. First, a machine learning
algorithm based on functions similar to those of the Haar
and Ada-Boost algorithms is used to train the classifier to
detect vehicles in the input image, which allows recognition
of the image of interest (ROI). Then additional training is
carried out using the basic component analyzer to learn how
to recognize the different types of vehicle samples. Density
based traffic signal system.
ISO 9001:2008 Certified Journal
|
Page 142
International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019
p-ISSN: 2395-0072
www.irjet.net
[2] Traffic control at a roundabout. It discusses the
differences between system setup and vehicle recognition,
tracking and routing systems and all technologies based on
driving vision in a car. Recently, all directional cameras and
parameters set by man are used in roadside systems, but
they are faced with automation problems. Technologically
integrated vehicles use light detection and ranging systems,
as well as radar and stereo systems, which is a continuous
overview. In the recent side system literature, vehicle
tracking is primarily based on a rare feature combined with
the Canada-Lukas-Tomasi settlement algorithm. In addition,
relocation, modeling and prior knowledge are necessary for
the exact location of the site and the correct classification of
participants. When used in vehicles, the first goal is to
precede or close the vehicle. Traffic signs are mainly used for
various adaptations of the optical data stream. When a
vehicle is detected, it is usually followed by general Bayesian
algorithms, especially Kalman and particle filters. The latter
optimized optical catadioptics and arc optics for the entire
controlled vision is used.
[3] The main idea of the proposed system is the effective
use of observation systems using image processing methods.
The step of the intellectual surveillance system is video. To
record, extract frames from a video input, apply a Gaussian
blur using Gaussian blur, use a blur, also called Gaussian
smoothing. This is a Gaussian matrix calculation that is
smooth or blurry to reduce image noise. After applying the
Gauss function, the system is divided into two parts:
recognition of abandoned objects and recognition of a
person. When detecting abandoned objects, suspicious
objects are recognized by extracting the background,
determining thresholds and clots. When human spots are
detected, red blue green transforms into light and color
measurements and thresholds. The main and only drawback
of this system is the need to update the background when
changing the background.
[4] Implements an intelligent flow control system that allows
you to move ambulances. Each vehicle has a special RFID tag
(located in a strategic location) that prevents removal or
destruction. RFID readers, NSK EDK-125-TTL and
PIC16F877A systems are used to read RFID tags attached to
a vehicle. It calculates the number of vehicles traveling along
a given route for a specific period of time. It also determines
the network load and thus adjusts the green light time.
Automatic check of routes based on traffic density, allowed
the police effortlessly and reduced their overhead. Since the
entire system is automated, it requires much less human
intervention.
[5] Implements RFID (Radio Frequency Identification) and
Record Identification (NPR) systems for vehicle
identification and control. License Plate Identification (NPR)
is a truly embedded system that often recognizes a vehicle's
license plate. Intersection systems are used only to identify
the vehicle. The license plate (NPR) and control system are
identified by a combination of image processing and RFID
and are used to identify and authenticate the vehicle. Smart
© 2019, IRJET
|
Impact Factor value: 7.34
|
traffic system control and traffic avoidance system during
emergencies using Arduino and ZigBee 802.15.4 is used. The
selected text is then used for authentication, and the RFID
information is used for verification. Image recognition
methods, such as image capture, binary transformation,
segmentation, pattern generation, and matching, are used to
identify and control the vehicle. The work of both modules is
parallel. The challenge is to recognize the characters.
[6] Demonstrates machine aggregation of vehicle
characteristics (traffic) to control a system widely used to
prevent identification of a vehicle, ambulance and hijacked
vehicle. Each vehicle has a radio frequency identifier (RFID),
which identifies the signal of the vehicle aggregation
(traffic). The RFID reader expects vehicles to move along a
given route for a certain period of time. Depending on the
number of vehicles, if the vehicles (movement) are detected
by an infrared sensor, the green light comes on and the
vehicles spread evenly. If the RFID reader reads the stolen
vehicle, a police message is immediately sent to the control
room using the mobile subscriber module 300. In addition, if
the hospital car passes through ZigBee and Atmega328, as
indicated by the vehicle (traffic), the message will be
redirected to the vehicle's resources (traffic) so that the
controller turns green. The microcurrent Atmega328 is
directly connected to the individual vascular components
and acts as a central point.
[7] It presents three main objectives and notes the speed of
the vehicle, marking them with paint and creating an
economical and simple system. In this case, the ultrasonic
sensor detects temporary vehicles and sends data to the
Arduino. He then examines the Arduino and responds
through a relay mechanism to a gun equipped with a
solenoid valve. Arduino provides programmed more
accurate results. It also states that the prototype system can
effectively detect and signal rapid acceleration. The result
shows that this innovative technology improves the
reliability of the road safety system. By performing these
functions in real time, you can avoid accidents up to 70%.
[8] Represents a system that identifies a unique RFID tag for
each vehicle, and each vehicle's identification data is entered
into the police database. The RFID reader reads the tag
values, and this value is provided by the Raspberry Pi
processor. RFID can be provided to hijack a car by selecting a
browser that serves to enter the Raspberry Pi. Raspberry Pi
compares the details of already saved RFID tags with
dynamic tags. When they match, they take a snapshot of the
vehicle and send it to a specific mailbox. The camera is
connected to the Raspberry Pi processor. The Raspberry Pi
processor saves image data on an SD card (hard disk). It uses
the Internet of Things and RFID, which connect various
smart devices that provide security when stealing a car.
ISO 9001:2008 Certified Journal
|
Page 143
International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019
p-ISSN: 2395-0072
www.irjet.net
3. EXISTING SYSTEM
The traffic lights system utilized in India are essentially preplanned wherein the season of every path to have a green
signal or light is settled. In a four-path traffic signal, one path
is given a green signal at any given moment. Along these
lines, the traffic light enables the vehicles of all paths to go in
a grouping. In this way, the activity can progress either
straight way or turn by 90 degrees. So regardless of whether
the activity thickness in a specific path is the minimum, it
needs to sit tight superfluously for quite a while and when it
gets the green flag it pointlessly makes different paths sit
tight for significantly longer lengths. Numerous techniques
had me acquainted with take care of the issue of activity
utilizing sensor and fluffy rationale strategies, but the issue
constant illuminating the issues is still tested. This issues can
overcome by utilizing Digital Signal Processing Technique i.e.
image Processing.
4. PROPOSED SYSTEM
Thingspeak Application: Sending notification with
embedded system was a problem with few solutions,
widespread was SMS but other side there are some issues of
extra payment, limited number of data, etc. Thingspeak
provide push notification from cloud after getting registered
on thingspeak website. In this system, the image is captured
via USB camera, and programming is performed in the
Embedded C language and Arduino Integrated Development
Language. Whenever an RFID sensor detects a signal, it
sends a positive signal to the Arduino. The RFID recognizes
the signal and sends the output signal to the USB camera to
capture the image.
Block diagram for traffic light system for the real time traffic
management system which overcome the drawback of
previous standard methods. The vehicle is detected by the
framework using image instead of sensor. A camera is placed
apace with the traffic light. It takes the image or captures the
image. To control the change of traffic lights, technique of
image processing is used. Through this amazing technique,
the traffic will decrease and abstain from the amount of time
being deteriorated by green light signal on the vacant road.
In proposed system emergency vehicle is detect on any rout
that time system send the signal to controller to clear that
road or traffic i.e., green signal is gives to that road. This
system is also more efficient and reliable in assessing the
presence of vehicles on the road, since it uses a system of
actual traffic density. Hence with the additional of these
techniques proposed system is more effective that existing
system.
5. METHODOLOGY:
5.1 Image acquisition
Fig -1: System Block Diagram
The functions of the various components are given below:
USB Camera: USB Camera captures the image and sends it
to the USB port of the Arduino Uno board. The camera model
used here is USB Camera model 5.0.
Arduino: Arduino is an open source platform used to
Design electronics. Arduino embodies a physical board
(often called a microcontroller) that is programmable and a
piece of software or integrated development environment
(IDE) used on computer to write and download computer
code on a physical board.
RFID Reader: An RFID reader (RFID Reader) is a device
used to collect information from an RFID tag used to track
individual objects. Radio waves are used to transmit data
from the tag to the reader.
RFID Tag: An RFID tag is an electronic tag that
communicates with an RFID reader via radio waves.
© 2019, IRJET
|
Impact Factor value: 7.34
|
In general, an image (image) is a two-dimensional
function f (x, y) (here x and y are flat coordinates). In
addition, it is also called gray level image (s). These x and y
values need to be changed to limited discrete properties to
create a digital image. The input image is a database from a
stunning database and drive database. An image of the retina
is made to prepare and monitor the condition of the person.
For processing on a PC, the analog image must be converted
to a digital image. All extended images made from limited
components and each limited component are called pixels.
5.2 Image cropping
The second step is to select a focused area, structuring
the cut calculations in MATLAB. The motivations for editing
is to recognize the street (path) where there are vehicles and
prevent meaningless source data. This meaningless data is
located in each frame of captured images.
5.3 Image matching
Recognition procedures depend on the compliance of
the sample vector of the sample corresponding to each class.
An indefinite example refers to a class that is closest to a
predefined metric. The least complex methodology is the
ISO 9001:2008 Certified Journal
|
Page 144
International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019
p-ISSN: 2395-0072
www.irjet.net
main classifier-separator, which, as its name implies,
excludes computers (Euclidean) between the blurry and all
model vectors. You choose the shortest distance for your
choice. There is another relationship-dependent
methodology that can be directly identified with images and
is very instinctive.
5.4 Image scaling
5.9 Traffic density
The next step is to calculate the mass of the car on the
road in the target zone. The signal is triggered by traffic
jams. The system changes the signal when there is a large
amount of car on the tape. The system captures images from
the tape and makes a decision to change the traffic signal
depending on the traffic density on the tape or path.
Image scaling occurs in all computerized photos at some
stage, regardless of whether it is in the Bayes demo tape or
in the design of photos. This happens whenever you change
the image size from one pixel to another. Image resizing is
fundamental when you need to increase or decrease the total
number of pixels. Regardless of whether a similar image size
is reproduced, the result may vary significantly, relying on
the calculation.
5.10 Background Subtraction
5.5 Image enhancement
In SVM algorithm, find the points closest to the lines of
both classes, which are called support vectors. Now calculate
the distance between the line and the support vectors. This
calculated distance is the margin. Main aim is to maximize
the difference. The optimal hyper plane is the one in which
this margin is maximum. SVM is therefore trying to make the
decision boundary so that the distance between the two
classes is outspread.
Image enhancement is a way to modify advanced images
to make the results more suitable for exposure or further
study. For example, we can reject noise, which will make less
demanding recognition of key qualities. On bad images of
differentiation, nearby characters melt in the midst of a
binarization and it is necessary to reduce the scatter of
characters on which previously imposed restrictions on the
word "picture".
5.6 Color detection
The idea of defining color, as the name itself recommends,
is part of image processing, which involves the separation
between elements depending on their color. If we work with
an image of different colors and we need to process only
certain colors, at this point the color matching strategies
essentially return a parallel image, where only the bits with
the important color are white and the output color is black.
This reduces the image data only to raster images, which
makes them less demanding for processing for various tasks.
Background subtraction is a widely used approach for
detecting moving object video from still cameras. The reason
for this approach is to detect moving objects of the difference
between the current frame and the reference frame, often
referred to as the background image or background model.
5.11 Support Vector Machine
6. WORKING MODULE
1. Arduino UNO, which controls the camera to capture all or
part of the image strip, is used. Recorded images will be sent
to MATLAB for processing.
2. MATLAB stops image processing, and the priority of each
band is to determine traffic density. Traffic density must be
determined for each IN road section.
3. A line or path with a higher traffic density receives a first
priority, and a path with a lower traffic density is the lowest
priority.
4. The road is selected in order of decreasing priorities.
5.7 Blob detection
Blob detection methods have dematerialized to discern
areas in a digital image that differ in properties, such as
luminosity or color, in contrast to the concealing areas.
There is a certain range, which shows that the protest will be
additionally checked if it exists in the range, or will be
considered a pictorial image and pass through it.
5. The time of each signal depends on the OUT bands or
traffic density in descending order.
6. When all the lane or route has given the green signal based
on their priority the traffic system complete its one cycle.
This process will be repeats and time for all signals will be
given on the basis on traffic density.
5.8 Object counting
To read the elements in the picture, the nearby borders of
the articles are different. The outer boundaries of the object,
in addition, the open borders within these products in a
parallel image are classified to distinguish between vehicles
that are available in a focused area.
© 2019, IRJET
|
Impact Factor value: 7.34
|
ISO 9001:2008 Certified Journal
|
Page 145
International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019
p-ISSN: 2395-0072
www.irjet.net
7. PROPOSED ALGORITHM
9. RESULTS
Table -1: Signal Processing
Initially, all signals at a particular junction will turn yellow
and as the traffic increases the signal will start turning green
in sequence. The normal flow of traffic shows that signal 1
will turn green first, and then signal 2 followed by signal 3
and at last signal 4. The order of the traffic lights began as a
green light that lights up at traffic signal 1, and the red light
of other traffic lights. Each vehicle has been given 2 seconds
time to pass through the particular signal. The duration of
this mode lasted a few seconds and it depends on the camera
image capture.
Then the sequence turned on the green light of traffic light 2
and the red light of other traffic lights for few seconds. The
same thing happened with traffic light 3 and traffic light 4
after the yellow lights of each traffic light turned on for a few
seconds. The microcontroller constantly repeats this light, if
the emergency order is not activated.
Fig -2: Traffic Signal System Algorithm
8. MATHEMATICAL MODEL
1. Input Vehicle Count
I = {i1, i2, i3, ...in| Where I is the vehicle count’
} i1,i2, i3,.in are the words of string.
2. Identify count on signal
S = {s1, s2, s3, ...sn | Where, S is the main set of counts.’ }
3. Identify features
F = {f1, f2, f3, fn | Where, F is main set of
features(Emergency vehicle /Normal Vehicle’ }
4. Identify Classification
C = {c1, c2, c3, fn | Where, F is main set of emergency
vehicle Classification depend on signal’ }
Classification of Emergency or Normal vehicle
If f1∈ Σ c1,c2….cn
C
Chart -1: Vehicle Priority, Emergency and Count Results
f1
In the graph above, there are three sectors available
according to the number of vehicles. In the above pictorial
representation, the maximum number of vehicles is six and
minimum number of vehicles is zero. Zero to six numbers
also represents the count of vehicles and 0-2 represents low
priority, 2-4 represents medium priority and 4-6 represents
high priority. So at signal 1, the count of vehicles is 2, at
signal 2, count of vehicles are 6 and at signal 3, the count of
vehicles is 3. In this case, signal 2 has maximum number of
vehicles so there is traffic congestion and so the signal for
signal 2 will turn green and all other signals will turn red.
Also, signal 4 is considered a road with no vehicles, if there is
more traffic then vehicles will go through that signal or path
to make way for emergency vehicle and after that it comes
.
.
If fn ∈ Σ c1,c2….cn
C
fn
Where f1,f2 = extracted feature
C= classification group
Calculating Time
Pi=fi/Fx
Where
Pi=polarity of vehicle
fi=feature of vehicle(emergency/normal vehicle)
Fx=total number of features
© 2019, IRJET
|
Impact Factor value: 7.34
|
ISO 9001:2008 Certified Journal
|
Page 146
International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 06 Issue: 11 | Nov 2019
p-ISSN: 2395-0072
www.irjet.net
back to normal. RFID is used to detect vehicle and keep a
record of the passing vehicle.
[10]
10. CONCLUSION
This paper presents a strategy for evaluating traffic systems
using image processing. This is eliminated when using
images taken from a carriage or tape, and recorded images
are moved into a series of images. Each picture is prepared
independently and the number of the vehicle is collected. If
there is no chance that the number of vehicles will exceed
the established limit, a notice of significant movement will be
indicated automatically. In the proposed system, the key
feature is ambulance priority. The advantages of this new
method include such advantages as the use of image
preparation in comparison with sensors, simplicity, and ease
of setup and, as a rule, high accuracy, low price and speed.
Since this method is implemented using MATLAB image
processing and programming, the cost of production is low,
with high speed and better performance for accuracy.
[11]
[12]
[13]
[14]
[15]
Technology and Exploring Engineering (IJITEE),
February 2019.
Elizabeth Basil, Prof. S. D. Sawant, “IoT based Traffic
Light Control System using Raspberry Pi” Department of
Electronics and Telecommunication NBNSSCOE, IEEE
2017.
Madhavi Arora, V. K. Banga, “Real Time Traffic Light
Control System”, 2nd International Conference on
Electrical, Electronics and Civil Engineering
(ICEECE'2012), pp. 172-176, Singapore, April 28-29,
2012.
Vikramaditya Dangi, Amol Parab, Kshitij Pawar & S.S
Rathod, “Image Processing Based Intelligent Traffic
Controller”, Undergraduate Academic Research Journal
(UARJ), Vol.1, Issue 1, 2012.
Bilal Ghazal, Khaled ElKhatib, Khaled Chahine, Mohamad
Kherfan “Smart Traffic Light Control System” 2016 IEEE.
Intelligent Traffic Light Control System using Image
Processing, Fr. C Rodrigues Institute of Technology,
Vashi.
http://www.slideshare.net/louiseantonio58/imageprocessing-based-intelligent-traffic-controlsystemmatlab-gui.
REFERENCES
[1]
[2]
[3]
[4]
[5]
Yong Tang & Congzhe Zhang & Renshu Gu & Peng Li &
Bin Yang, “Vehicle detection and recognition for
intelligent traffic surveillance system,” Springer Science +
Business Media New York 2015.
Sokèmi René Emmanuel Datondji, Yohan Dupuis, Peggy
Subirats, and Pascal Vasseur, “A Survey of Vision-Based
Traffic Monitoring of Road Intersections,” in IEEE
transactions on intelligent transportation systems, 2016.
Kitae Kim, Slobodan Gutesa, Branislav Dimitrijevic,
Joyoung Lee, Lazar Spasovic, Wasif Mirza and Jeevanjot
Singh, “Performance Evaluation of Video Analytics for
Traffic Incident Detection and Vehicle Counts
Collection,” Springer International Publishing AG, 2017.
S. Sarkar and C. Parnin, “Characterizing and predicting
mental fatigue during programming tasks,” in
Proceedings of the 2nd International Workshop on
Emotion Awareness in Software Engineering. IEEE
Press, 2017.
Rajeshwari Sundar, Santhoshs Hebbar, and Varaprasad
Golla, “Implementing Intelligent Traffic Control System
for Congestion Control, Ambulance Clearance, and
Stolen Vehicle Detection,” IEEE SENSORS JOURNAL, VOL.
15, NO. 2, FEBRUARY 2015.
Aqib Mehmood, Mumtaz Ali, Rameez Ahmad, Sayed
Irfan, Habib ur Rahman, “Identification and verification
of vehicle using rfid technique,” VAWKUM Transactions
on Computer Sciences, 2016.
[7] Sivakumar.R, Vignesh.G, Vishal Narayanan, Prakash.S,
Sivakumar. V, “Automated Traffic Light Control System
and Stolen Vehicle Detection,” IEEE International
Conference on Recent Trends in Electronics Information
Communication Technology, 2016.
[8] Vasanth B, Sreenivasan S, Mathanesh V.R, “Over Speed
Vehicle Marking System Using Arduino UNO Controlled
Air Cannon,” International Journal of Engineering
Technology Science and Research, September 2017.
[9] P. Devika, V.Prashanthi, G.Vijay Kanth, J Thirupathi,
“RFID Based Theft Detection and Vehicle Monitoring
System using Cloud,” International Journal of Innovative
[6]
© 2019, IRJET
|
Impact Factor value: 7.34
|
ISO 9001:2008 Certified Journal
|
Page 147